Parental Divorce in Australia, Cohorts Born 1900 - 1975

Melbourne Institute Working Paper No. 22/04

Date: September 2004


M. D. R. Evans
Jonathan Kelley


This paper examines the sources of parental divorce in Australia using respondents' retrospective reports of parents' behaviour in the IsssA-Pool database, a pooled series of representative national samples of Australia conducted between 1984 and 2002 (N=19,601 valid cases for this analysis). We analyse the probability of divorce using logistic regression models. The results include a very large effect of time, with people reaching age 14 before the Family Law Act took effect being much less likely to grow up in divorced families, even net of a wide variety of other social and economic changes over time. Important also are several indicators of family traditionalism: Parents from the Mediterranean countries and parents who are faithful church-goers have more stable marriages. The effects of gender and of size of place are ambiguous: they are marginally significant statistically, and small, to boot. Parental social class has little, if any, effect, except that parents at the top and bottom ends of the income distribution are probably less likely than those in the middle to divorce. Family size and religious denomination do not have significant effects on the probability of divorce in these models. Finally, there is a very large connection between maternal employment and parental divorce; we cannot disentangle the causality here, but can establish a (large) upper bound on the effect of maternal employment on divorce.

Download Paper